Deep Learning Transformer Architecture for Named Entity Recognition on Low Resourced Languages: State of the art results

TL;DR

This study evaluates transformer-based deep learning models for Named Entity Recognition on ten low-resourced South African languages, demonstrating significant performance improvements over traditional models and highlighting their practical NLP applications.

Contribution

The paper presents the first comprehensive evaluation of transformer models for NER on low-resourced South African languages, showing their superiority over other neural and machine learning models.

Findings

Transformer models achieved the highest F-scores for six of ten languages.

Fine-tuning transformer models per language enhances performance.

Transformer models outperformed Conditional Random Fields in average F-score.

Abstract

This paper reports on the evaluation of Deep Learning (DL) transformer architecture models for Named-Entity Recognition (NER) on ten low-resourced South African (SA) languages. In addition, these DL transformer models were compared to other Neural Network and Machine Learning (ML) NER models. The findings show that transformer models substantially improve performance when applying discrete fine-tuning parameters per language. Furthermore, fine-tuned transformer models outperform other neural network and machine learning models on NER with the low-resourced SA languages. For example, the transformer models obtained the highest F-scores for six of the ten SA languages and the highest average F-score surpassing the Conditional Random Fields ML model. Practical implications include developing high-performance NER capability with less effort and resource costs, potentially improving downstream NLP tasks such as Machine Translation (MT). Therefore, the application of DL transformer architecture models for NLP NER sequence tagging tasks on low-resourced SA languages is viable. Additional research could evaluate the more recent transformer architecture models on other Natural Language Processing tasks and applications, such as Phrase chunking, MT, and Part-of-Speech tagging.